Based upon the strong statistical association between aerobic capacity and all-cause morbidity and mortality, it was hypothesized that artificial selection of rats for low and high aerobic exercise capacity would yield models that also contrast for disease risks. If true, this would support the notion that impaired oxygen metabolism is a common feature that mechanistically underlies disease risks. Twenty generations of bidirectional selection produced lines of low capacity runners (LCR) and high capacity runners (HCR) that differ by over 5-fold in aerobic treadmill running capacity. The LCR score high on numerous risks including the metabolic syndrome and the HCR score high for health factors such as maximal oxygen consumption. Importantly, the LCR also respond more to environmental health risks such high fat diet. The long-term goals are to understand mechanistically the features that divide these contrasting models for disease risk and to ultimately make these models an economically viable and self-sustaining resource. The specific aims are to: 1) continue two-way artificial selection until divergence plateaus, 2) generate and phenotype LCR and HCR for mechanistic evaluation of physiologic and disease risk differences;special emphasis will be given to evaluation of differential gene expression using novel bioinformatics approaches, and, 3) stabilize genotypes within the models by producing a panel of inbred strains that represent the 5-fold divide for running capacity. We describe four large-scale studies that are being pursued aggressively as representative examples: a) mechanistic evaluation of differential for aging and longevity between LCR and HCR (Russ Hepple), b) test if exercise or caloric restriction in very young LCR or HCR has a long-term positive "imprinting" effect on metabolic endpoints (Charles Burant), c) interpret the large number of expression arrays being generated with a "molecular concept map" that navigates 15 databases (Arul Chinnaiyan), and 4) explore the molecular basis for higher anxiety and depression in the LCR compared to HCR (Huda Akil). The genetic and environmental variation of human populations imparts considerable difficulty to the study of complex diseases, making animal models an attractive path. The LCR and HCR are hypothesis-based models that are currently known to divide for numerous disease risks, with more likely to be defined. Information obtained from these models will immediately suggest pathways for translational studies for more effective modes of diagnosis, prevention, and treatment of complex diseases.